Iowahawk (yes, that Iowahawk) provides a step-by-step manual for the layman to recreate, or at least understand in some detail, the statistical methodologies involved in gathering climate change data. Very interesting. We’ll skip to the discussion segment:

all that math-y spreadsheet-y stuff above was not meant to perfectly replicate any specific study done by Mann et al.; those specific studies differ by the choice of instrumental temperature data set, the choice of proxy variables, whether series are smoothed with a filter (Fourier transform etc), and so on. My goal was to provide interested people with a hands-on DIY example of the basic statistical methodology underlying temperature reconstruction, at least as practiced by the leading lights of “Climate Science.”

If you’ve followed all this, it should also give you the important glossary terms that should help you decipher the Climategate emails and methodology discussions. For example “instrumental data” means observed temperature; “reconstructions” are the modeled temperatures from the past; “proxy” means the tree ring, ice core, etc. predictors; “PCs” mean the principal components.

Is there anything wrong with this methodology? Not in principle. In fact there’s a lot to recommend it. There’s a strong reason to believe that high resolution proxy variables like tree rings and ice core o-18 are related to temperature. At the very least it’s a more mathematically rigorous approach than the earlier methods for climate reconstruction, which is probably why the hockey stick / AGW conclusion received a lot of endorsements from academic High Society (including the American Statistical Association).

The devil, as they say is in the details. In each of the steps there is some leeway for, shall we say, intervention. The early criticisms of Mann et al.’s analyses were confined to relatively minor points about the presence of autocorrelated errors, linear specification, etc. But a funny thing happened on the way to Copenhagen: a couple of Canadian researchers, McIntyre and McKitrick, found that when they ran simulations of “red noise” random principle components data into Mann’s reconstruction model, 99% of the time it produced the same hockey stick pattern. They attributed this to Mann’s method / time frame for selecting of principle components.

To illustrate the nature of that debate through the spreadsheet, try some of the following tests:

Run step 3 through step 7, but only use the proxy data up through 1960 instead of 1980.

Run step 5 through step 7, but only include the first 2 principle components in the regression.

Run step 3 through step 7, but delete the ice core data from the proxy set.

Run step 2 through step 7, but pick out a different proxy data set from NOAA.

Or combinations thereof. What you’ll find is that contrary to Mann’s assertion that the hockey stick is “robust,” you’ll find that the reconstructions tend to be sensitive to the data selection. M&M found, for example, that temperature reconstructions for the 1400s were higher or lower than today, depending on whether bristlecone pine tree rings were included in the proxies.

What the leaked emails reveal, among other things, is some of that bit of principal component sausage making. But more disturbing, they reveal that the actual data going into the reconstruction model — the instrumental temperature data and the proxy variables themselves — were rife for manipulation. In the laughable euphemism of Philip Jones, “value added homogenized data.”

One way or another, the truth looks likely to emerge rather definitively over the next several years about what is known and what is speculation. Good. HT: Ace

This entry was posted on Sunday, December 13th, 2009 at 6:57 am and is filed under business, Democrats, General, Republicans, Science. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.